OpenAI Rolls Out New Advanced Voice Mode for ChatGPT

As a seasoned tech analyst with over two decades of experience in the industry, I have seen AI evolve from a mere concept to a significant part of our daily lives. The latest development by OpenAI, the rollout of advanced Voice Mode for ChatGPT Johansson, is undeniably impressive.


ChatGPT Johansson, powered by OpenAI, is now offering an enhanced, realistic voice feature to a select number of ChatGPT Plus members. This means these users can enjoy responses that sound remarkably human-like for the very first time.

We’re gradually making available an enhanced conversation feature called “Advanced Voice Interaction” to some ChatGPT Plus subscribers. This new mode aims for more lifelike, immediate discussions, lets you interject at any point, and is designed to understand and react to your feelings in real-time.

— OpenAI (@OpenAI) July 30, 2024

At the unveiling of GPT-40 in May, this characteristic was first showcased, attracting notice for its advanced functions. Nevertheless, it encountered criticism because its voice resembled Scarlett Johansson’s closely, sparking discussions about ethics and law due to these similarities.

At OpenAI’s event, the latest version of the speech mode displayed significant improvements compared to its predecessor, demonstrating the ability to adapt dynamically to interruptions and adjust its course when needed. The staff at OpenAI showcased this functionality by having the chatbot interact in real-time.

In a surprising turn of events, the stage character, nicknamed “Sky,” faced criticism even after enhancements due to its striking similarity to Scarlett Johansson’s portrayal of an AI in Her. Consequently, Johansson reached out to OpenAI for additional details regarding the voice’s origin.

Initially planned for a beta launch in late June, the rollout got pushed back by a month due to OpenAI’s need to meet their security standards and improve the model’s ability to filter specific types of data.

According to OpenAI’s representative, Taya Christianson, the speech model was subjected to rigorous testing by over a hundred external experts, often referred to as “red teamers,” whose job is to exploit potential weaknesses in technology. The decision to postpone the release was a response from OpenAI to the increased attention on its safety measures.

In their latest update, OpenAI introduced filters in the new voice setting to prohibit requests for creating music or copyrighted sounds. Following concerns about a voice similar to Scarlett Johansson, OpenAI has confined this mode to just four pre-recorded voices, voiced by professional actors. Taya Christianson, a representative from OpenAI, confirmed that ChatGPT won’t mimic other people’s voices. Any generated output that deviates from these preset voices will be prevented to avoid misuse.

By autumn, OpenAI intends to grant access to the enhanced voice feature for all ChatGPT Plus subscribers. The intention behind this release is to offer a more engaging and adaptable user experience, all the while ensuring top-notch safety and ethical practices remain intact.

Read More

2024-07-30 23:17